Drift on ocotomap and robot wheel odometry
Hi,
Octomap has lots of drift only using robot odometry, shall the localization(such as amcl) be used at the same time?
ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange |
Hi,
Octomap has lots of drift only using robot odometry, shall the localization(such as amcl) be used at the same time?
octomap is a representation of input (3d data with transformations between them). It is the responsibility of the user to provide accurate transformations between two point clouds to create accurate 3D octomap based representations. Definitely one has to use some kind of localizations (amcl in 2D) and RGB-D SLAM (for example in 3D) to provide accurate transformations between two point clouds. Make sure that the algorithms perform loop closure, or else it may not give good 3D representations.
All type of odometry (either wheel or visual) accumulate errors over time and hence places need to be revisited, detected and accumulated error need to be corrected. This is generally done in a graph optimization framework after detecting the loop closures.
Good luck
Asked: 2014-10-31 20:03:49 -0500
Seen: 417 times
Last updated: Nov 01 '14
robot_localization Map, Odom, and drift.
using existing map instead of slam-built map to navigation and localization
Can the robot_localization package be used with GPS-only?
gmapping make missing map when follows a straight wall
amcl does not publish transforms to "future"
[Autoware] Lidar fake perception
Writing a cmd_vel node with amcl_pose to move robot